Understanding AI modes: In-Depth, Fast, Plan, and Outcome
Andri answers in one of four modes, and you choose the mode before you send a message: In-Depth Mode, Fast Mode, Plan Mode and Outcome Mode. In-Depth Mode is the default and handles most work.
Switching modes
The mode button in the message composer always shows the mode you are in. Click it and pick another mode from the list.
Fast Mode stays on until you turn it off. Plan Mode and Outcome Mode apply to the message you send next: once that work is done, the chat returns to In-Depth Mode, so set them again for the next message that needs them.
In-Depth Mode
In-Depth Mode is the default. Andri researches your case documents, your company knowledge base and the legal sources for the jurisdiction of the case, then answers in a single reply. Use it for legal research, precedents, analysis, and anything that needs citations.
Fast Mode
Fast Mode returns short, quick answers to simple questions, such as checking a date in a contract.
Fast Mode is not available for messages that carry @ references. If you add a document, tag or action to your message, Andri returns to In-Depth Mode.
Plan Mode
In Plan Mode, Andri drafts a structured plan and does none of the work until you approve it. The plan sets out what Andri understands, the deliverables, the success criteria, and what it treats as out of scope.
Review the plan, then choose Execute Plan to start the work, Edit Plan to refine it by typing your changes in the chat, or Cancel Plan to drop it.
Outcome Mode
Outcome Mode is for large, graded deliverables. Andri builds a rubric of criteria together with you, then keeps working until every criterion is met.
The work runs as a background job, so you can leave the chat and follow its progress while it continues. Another thread in the same case gives you somewhere to work in the meantime. Choosing "Turn it into a graded task" while improving a prompt also puts you in Outcome Mode.
Choosing the right mode
Stay in In-Depth Mode for everyday questions and research. Switch to Fast Mode when you want a single fact back quickly, to Plan Mode when the work runs over several steps and you want to see the approach before Andri commits to it, and to Outcome Mode for a long deliverable that should be graded against criteria you agree first.
Mode affects processing time and token usage. In-Depth Mode, Plan Mode and Outcome Mode do more work and use more tokens than Fast Mode.